S UBCORTICAL PATHOLOGY AS A FUNCTION OF NEUROPSYCHOLOGICAL DEFICITS

The results of Study 3 provide evidence of increasing subcortical gray matter pathology along the cognitive phenotypes of ALS, ALS-Plus, and ALS-FTD, suggesting that the functional continuum proposed by neuropsychological studies is driven by an underlying pathological continuum. Moreover, vertex analyses in ALS-FTD revealed preferential atrophy of sub-regions within the basal ganglia, which connect to established cortical sites of ALS pathology and suggest a network-wise vulnerability of interconnected gray matter regions.

Our results indicate that hippocampal and thalamic volume reductions are the most sensitive predictors of cognitive phenotype classification. As pointed out in section 6.2, hippocampal pathology is a well-documented feature of ALS (Takeda et al., 2009) and corresponds with stage 4 of the recently proposed pTDP-43 pathological staging system(Brettschneider et al., 2013). Hippocampal atrophy has been previously linked to memory deficits in ALS (Abdulla et al., 2014), and hippocampal dysfunction seems to manifest later than motor dysfunction (Stoppel et al., 2014). Here, we confirm both volumetric and vertex-wise hippocampal changes, and demonstrate that hippocampal atrophy is closely associated with verbal memory performance. While ALS patients without cognitive impairment showed no significant

General discussion hippocampal changes, ALS-Plus patients exhibited volume loss in the head of the hippocampi. A similar anatomical pattern was observed in ALS-FTD, but with a higher degree of volume loss, and surface-projected changes expand from the head of the hippocampus to the body of the structure. Our findings underscore the importance of cognitive assessment beyond executive function in ALS.

Thalamic atrophy was identified as another sensitive discriminator of cognitive categorization, and vertex analyses revealed a distinct pattern of thalamic involvement in ALS-FTD. These changes preferentially affect the anterior nuclei, which have projections to the limbic system, especially to the hippocampus and cingulate gyrus. This anatomical vulnerability corresponds with the prominent behavioral deficits observed in ALS-FTD.

Additional atrophy was observed in the anterior ventral and lateral thalamic nuclei, which are known to project to motor and supplementary motor regions. Motor cortex pathology is a hallmark feature of ALS (Bede et al., 2013a); therefore the degeneration of associated thalamic nuclei is not surprising. These observations warrant further studies of thalamocortical connectivity to explore its contribution to motor planning and disability in ALS.

Caudate nucleus pathology in ALS-FTD affected both the head and tail of the structure. The head of the caudate receives input from the dorsolateral prefrontal cortex(Alexander et al., 1986), a region directly linked to executive function (Poston and Eidelberg, 2012). The tail of the caudate is associated with visuo-spatial information processing (Lawrence et al., 2000), which is known to be impaired in ALS(Phukan et al., 2012) and deteriorates with disease progression(Elamin et al., 2013). Moreover, the caudate nucleus is strongly interconnected with the putamen, where, in turn, dorsal atrophy was also observed. Dorsal putaminal regions receive afferentation from the motor-, premotor-, and supplementary motor cortices, all of which are heavily affected in ALS (Bede et al., 2013a). Interestingly, other studies of subcortical pathology in ALS did not capture changes in the putamen (Bede et al., 2013c;

Westeneng et al., 2014), which is likely to be explained by their unsegregated patient cohorts.

In addition to dorsolateral- and motor-related frontostriatal pathology, considerable changes were also identified in the accumbens nuclei, a structure that primarily receives input from the anterior cingulate cortex (Alexander et al., 1986). Accumbens pathology in ALS has been previously linked to the C9orf72 genotype (Bede et al., 2013c) and used as a stage-defining

General discussion nucleus accumbens is widely interconnected with the ventromedial prefrontal cortex (O'Callaghan et al., 2014), mediates motivation and plays a key role in reward-seeking behavior(Elamin et al., 2013). Consistent with its role in motivation, an association has been identified between accumbens atrophy and apathy measures.

Taken together, a distinctive pattern of pathology was observed within basal ganglia sub-regions, which are directly connected to the prefrontal, motor, and orbitofrontal cortex, i.e., key cortical regions of ALS pathology. The identified anatomical vulnerability of interconnected cortical and subcortical gray matter regions further supports the concept of network degeneration in ALS. Importantly, the observed imaging changes were independent from both disease duration and motor disability (ALSFRS-R) and are not merely a function of prolonged disease duration or higher disease burden.

Density analyses revealed cerebellar gray matter changes in C9orf72-negative ALS-Nci (Figure 10a). To date, cerebellar changes have been mostly associated with the C9orf72 hexanucleotide repeat expansion both in ALS and FTD(Mackenzie et al., 2014). However, recent imaging studies have reported white matter pathology in non-demented C9orf72-negative ALS patients (Bede et al., 2015) and cerebellar gray matter atrophy along the ALS-FTD spectrum(Tan et al., 2014). Our whole-brain analyses highlight widespread cortical and basal ganglia pathology in ALS-FTD, not only in comparison to controls but also to ALS-Nci and ALS-Plus patients. While the extent of cortical gray matter atrophy in ALS-FTD (Figure 10b/c) is consistent with previous reports (Mioshi et al., 2013), basal ganglia involvement has not been previously characterized in this phenotype. We have identified considerable gray matter density reductions in the thalamus, caudate, putamen, and pallidum in ALS-FTD complementing our volumetric and vertex findings. These results are consistent with an imaging study of FTD without ALS (Garibotto et al., 2011).

One of the limitations of our study is the relatively small numbers of ALS-FTD and ALS-Plus patients. However, the observed anatomical changes were consistent across multiple imaging methods and demonstrated in volumetric, shape and density analyses. There is relatively little known about C9orf72-negative ALS-FTD, as landmark imaging studies of ALS-FTD(Chang et al., 2005) predate the discovery of C9orf72 and are likely to have used admixed cohorts of hexanucleotide carriers and non-carriers. Studies published after the identification of C9orf72 repeats on the other hand focus mainly on the imaging signature of C9orf72. Our study

General discussion characterizes a relatively small, but unique cohort of C9orf72-negative ALS-FTD patients.

Another study limitation may stem from classifying patients into ALS-ci and ALS-bi based on the Strong criteria(Strong et al., 2009), which puts the emphasis on executive function, apathy and disinhibition for patient categorization. Recent reports, however, indicate that other cognitive domains, such as language(Abrahams, 2013), memory (Machts et al., 2014), and social cognition (Girardi et al., 2011) may also be pertinent in accurate patient categorization.

Our finding of hippocampus mediated memory performance supports these observations and needs to be considered when revising the current consensus criteria.

The current findings indicate that the cognitive phenotypes of ALS are associated with incremental subcortical gray matter pathology that can be captured by volumetric, shape, and density measures in-vivo. The diagnoses of ALS-Nci, ALS-Plus, and ALS-FTD represent hierarchical categories along the same pathological continuum. The presented findings also provide evidence of a network-wise vulnerability of interconnected cortical and subcortical gray matter regions. Our results highlight that imaging studies of cognition in ALS and FTD need to assess basal ganglia integrity and not rely on cortical gray and subcortical white matter measures alone. The detailed characterization of basal ganglia pathology is pivotal to elucidate the frontostriatal changes underlying cognitive dysfunction in ALS.

6.4. Functional connectivity changes indicate widespread temporal lobe